It is not uncommon for companies that delivery online digital content to use experimentation in the context of data-driven decision making. A/B testing may be used to perform an experiment pertaining to delivery of content online, and to determine which variant of two variants of the experiment better accomplishes a particular goal of the experiment. Usually, in A/B testing, a group of online users is divided into a treatment group to receive the treatment variant of the experiment, and a control group to receive the control variant of the experiment. Based on the user responses received from the members of the treatment group and the control group, a conclusion regarding the goal of the experiment may be drawn.